Working Paper, University of Utah, Department of Economics 2009-03
The paper offers an alternative approach to analyzing stock market time series data. The purpose is to develop descriptive, more intuitive, and closer to reality analogs of the behavior of US stock market prices, as indexed by the S&P500 stock price index covering the period October 2003 to October 2008. One analog developed is the escalator principle and the blind man. The approach is to treat prices as a random and independent variable and use extreme value theory to judge probabilistically whether prices and their attributes are from an initial universe or whether there has been a regime change. The attributes include the level, first difference, second difference and third difference of the ordered price series. Various graphing tools are used, such as, probability paper and different specifications of exponential functions representing cumulative probability distributions. The argument is that traditional time-series analysis implies a given universe, usually normal with either a constant or time-dependent variance (or measureable risk) and consequently does not handle well uncertainty (nonmeasureable risk) due to regime changes. The analogs show the investor how to determine when a regime change has likely occurred.